English

An asymptotic upper bound on prime gaps

Number Theory 2015-10-08 v3 Mathematical Physics math.MP

Abstract

The Cram\'er-Granville conjecture is an upper bound on prime gaps, gn=pn+1pn<\cCramerlog2png_n = p_{n+1} - p_n < \cCramer \, \log^2 p_n for some constant \cCramer1\cCramer \geq 1. Using a formula of Selberg, we first prove the weaker summed version: n=1Ngn<n=1Nlog2pn\sum_{n=1}^N g_n < \sum_{n=1}^N \log^2 p_n. In the remainder of the paper we investigate which properties of the fluctuations \fluc(x)=π(x)\Li(x)\fluc (x) = \pi (x) - \Li(x) would imply the Cram\'er-Granville conjecture is true and present two such properties, one of which assumes the Riemann Hypothesis; however we are unable to prove these properties are indeed satisfied. We argue that the conjecture is related to the enormity of the Skewes number.

Keywords

Cite

@article{arxiv.1506.03359,
  title  = {An asymptotic upper bound on prime gaps},
  author = {André LeClair},
  journal= {arXiv preprint arXiv:1506.03359},
  year   = {2015}
}

Comments

The proof of the last theorem of the last version is incorrect, because the fluctuations were treated too smoothly. We could only replace it with a weaker result

R2 v1 2026-06-22T09:51:08.844Z